Perplexity launches AI-powered patent search

Perplexity has launched Perplexity Patents, a tool that lets you search for patents by asking questions in natural language, without learning commands or building complex keyword combinations.

The service is available worldwide as a beta and will be free for all users during this stage. Pro and Max subscribers will get more available queries and additional options for configuring the models.

Search patents as if you were having a conversation

Instead of writing a technical search query, you can ask questions such as “are there patents for AI language learning?” or “what are the key quantum computing patents since 2024?”. Perplexity interprets your intent and returns related documents, with direct links to the original patents and an integrated viewer.

The search also keeps the context between questions. You can ask it to compare two inventions, look more closely at a specific patent, or explore a line of research without starting over. The system can suggest related topics to continue the search.

This aims to address a common problem: traditional tools often depend on exact terms. If you search for “fitness trackers”, for example, a conventional search engine may leave out patents that refer to “step-counting watches” or “wearable health-monitoring devices”. Perplexity says it can connect those concepts even when they do not use the same words.

A tool for finding prior art

In intellectual property, prior art includes earlier inventions and documents that may be relevant when evaluating a new idea. Finding it matters to researchers, companies and patent professionals, but specialized databases are often expensive and require training.

Perplexity Patents breaks complex queries into several search tasks. It then checks a specialized patent index and uses the results to guide further searches. Depending on the query, answers may draw on dozens or hundreds of documents.

The company also wants to go beyond patents. When necessary, the system can search:

• Academic papers.
• Public software repositories.
• Blogs and other publications.
• Videos and sources where new technical ideas appear.

The logic is simple: an innovation may appear first in a paper, a software demonstration or a video, rather than in a patent application. Searching across several types of sources can help identify related trends and technologies.

What this changes for you

For an independent inventor, a startup or a student, the main difference is access. You no longer need to know a database's syntax to start investigating whether an idea has precedents or which companies are working in a specific area.

It can also help you prepare an initial overview before consulting a specialist. But you should verify the answers against the original documents, especially when a decision could affect a patent application, an investment or a legal dispute.

For now, Perplexity Patents is a beta. It will be important to watch how it handles differences between jurisdictions, the accuracy of its results and its ability to distinguish a superficial match from genuinely relevant prior art. Patent search is becoming more accessible, but expert review is still necessary when something important is at stake.